Channel describes all media through which a signal is transmitted from a transmitter to a receiver. The media include physical media, such as cable channel, optical channel and wireless channel. The wireless channel is random in many aspects and may produce signal magnitude, phase and frequency distortion and intersymbol interference which are great challenges to the design of the receiver. Therefore, wireless channel estimation and prediction are needed. In theory studies, a channel model statistic method is introduced to describe the influence that a channel has on signals. A channel model is built based on the attributes of signals in certain environment.
Channel estimation can be defined as the qualitative analysis of the influence that a physical channel has on an input signal, and is the mathematical expression of the influence that the channel has on the input signal. “Good” channel estimation indicates an estimation algorithm that minimum estimation errors. Channel estimation can be roughly divided into non-blind channel estimation, blind channel estimation and semi-blind channel estimation. In the non-blind channel estimation, pilots are adopted to obtain the information of channel at the pilot positions, and the information of the channel for transmission of all data will be obtained based on the information of the channel at the pilot positions. In the blind channel estimation, channel estimation is obtained by using a corresponding information processing technique and no pilot is needed. Compared with conventional non-blind channel estimation technique, the blind channel estimation technique significantly improves the transmission efficiency of the system. However, the convergence speed of the blind channel estimation algorithm is usually low, which hinders the application of the blind channel estimation technique in practical systems. The semi-blind channel estimation technique finds a middle way between data transmission efficiency and convergence speed, i.e., a way to obtain the information of the channel at all times with comparatively less training sequences.